The difference is not proven unless there is a control group not exposed to any other known harmful substance.
Actually; that would violate the rules for a double blind study. If you're trying to isolate whether or not smoking causes increased rates of cancer you WOULDN'T want to sit there and isolate your test group from other substances.
If you sit there and do a test where you do a test group of Smokers+Coal exhaust+car exhaust+industrial cleaners and a control group of none of that, you don't have the same amount of isolation. All you get is that lung cancer decreases if you removed TSmoke&coal&car&cleaners. We know pollution from coal power plants is bad for you because there are increased illness rates around coal plants, that decreases as exposure decreases(you move further away or at least upwind on average from the plant).
As long as the other exposures are similar in both the test and control groups, it evens out.
I could question ten thousand cancer patients and ask how many of them have ingested, just for the sake of arguement, an "average amount" of refined sugar. Or how many have mercury amalgam dental fillings.
This is closer to the way studies such as this go(I'm oversimplifying):
We ask 10k people, of say 70 years old. The two questions are: Do you smoke? and Have you had(or currently have) lung cancer?
Let's assume at the time of the study, 50% respond yes to the first question. We end up having a 50% smoking rate for 70 year olds.
For the second question, we get that 5 non-smokers got lung cancer, and 28 smokers did.
More realistically, we hit all age groups and get more data points to analyze to control for things like age.
And you'd likely get that, at least for lung cancer, that the rate of fillings or sugar are NOT higher than the general population. Once you have correlation, then you start investigating closer. While double-blind tests would be nice, ethics about human studies combined with the length of time and actual percentage rates make that impractical.
If more than x percent said yes, many said "alot" and a few said "none", I could say that it is obvious that these two things cause cancer. If it is lung cancer, and you are going to say it was smoking tobacco - you have to exclude those who use refined sugar and have mercury amalgam dental fillings.
You're missing a critical point: What's the rate of these two items in the non-cancer population? Nearly everybody is going to have a filling - so while there will be a correlation if you only look at cancer patients. Such correlation will disappear if you consider that nearly 100% of people period have fillings.
When you look at lung cancer and see that 90% of those with it smoked, and the percentage of smokers in the general population is only 50% - then you have an issue.
Smoking tobacco is not going to kill anyone in moderation. Smoking a few modern factory cigarettes with chemically treated tobacco, paper and filter probably won't cause your heart to fail later on. Smoke a few packs a day and it is likely you are asking for trouble.
I think I mentioned something like that - cigarettes are targeted more because people tend to use them more. The occasional cigar isn't too bad. Heck, the occasional cig isn't too bad, but how many 'occasional' cigarette smokers are there? Then again - who's to say that chemically treated tobacco isn't worse than untreated tobacco? We're also not talking about heart failure - we're talking about lung cancer.
Show me any study on smoking proving that it causes cancer, with a control group that eliminate the effects of refined sugar. Or airborn particulates of the type that come down from industrial plants, construction sites, etc.
Like I said, a real control group, especially one of adequate size, controls for all of those. You don't eliminate refined sugar, you simply make sure that there's not a bias in your study group that's independent of the the control factor. Plenty of smokers eat lots of sugar, as do plenty of non-smokers. Thus I have to ask you: Do you have any evidence that either smokers or non-smokers consume more sugar?
Wikipedia states that less than 10% of lung cancer cases are non-smokers.
Gallop reports smoking rate in 1944 at 41%
Wiki also states that there's a 20 year lag between smoking rates and lung cancer.
Let me ask you a question: What environmental or behavioral condition, other than smoking, would cause smokers to vastly over represent themselves for cases of lung cancer?